regularizer_l2 {keras3} | R Documentation |
A regularizer that applies a L2 regularization penalty.
Description
The L2 regularization penalty is computed as:
loss = l2 * reduce_sum(square(x))
L2 may be passed to a layer as a string identifier:
dense <- layer_dense(units = 3, kernel_regularizer='l2')
In this case, the default value used is l2=0.01
.
Usage
regularizer_l2(l2 = 0.01)
Arguments
l2 |
float, L2 regularization factor. |
Value
A Regularizer
instance that can be passed to layer constructors or
used as a standalone object.
See Also
Other regularizers:
regularizer_l1()
regularizer_l1_l2()
regularizer_orthogonal()
[Package keras3 version 1.1.0 Index]